Closed kexul closed 3 years ago
this is default practice because the equilibrated input is implicitly equilibrated at one lambda value (even though lambda-dependent potentials are not used during equilibration), typically lambda=0.0 (but that depends on how your put together the equilibration-->free energy pipeline). Changing lambda changes interaction with the system, and can introduce steric clashes (depending on the perturbation) that will blow up the MD integrator with high magnitude forces.
Therefore we recommend to energy minimise the input at each lambda value, and discard the beginning of the output trajectories (to allow for requilibration) before free energy estimation. Typically 15-50 ps seems sufficient but that could depend on the exact use case.
Thanks, that makes things much clearer now!
Dear BioSimSpace developers,
Recently I found that the default protocol generated by
BSS.FreeEnergy.Relative()
setminimise=True
insomd.cfg
. When I passed an already equilibrated system (by Gromacs) to it, the energy would drop a lot during minimization.I'm concerned that:
BSS.FreeEnergy.Relative()
, I performedMinimise->NVT equilibration->NPT equilibration
by Gromacs. The energy before minimization agrees well with the potential energy printed by Gromacs.Any help is appreciated! 🤗
Here is the full log of
somd-freenrg
. somd.txt